Expert question for the day

[Jeff Murray]

Who knows if they shot the interviews on a Canon 5D? Or maybe some other camera/workflow producing similar issues to those we face?

Check the moiré on Richard King's shirt. Love this short by the way - incredible effort going into the soundtrack.

http://www.vimeo.com/staffpicks#13396749

With regards

Jeff

[Liam Hall]

Looks like 5D to me. Though to be fair that short would have produced moiré on far more expensive cameras.

[Bill Davis]

Moiré has NOTHING to do with the "cost" of a camera.

The first place I saw major moiré was watching the old Johnny Carson show where his ties and occasionally his jackets would crawl all the hell all the time.

You think that NBC in equipping it's #1 cash cow over those decades used CHEAP studio cameras?

Hardly.

The only issue is that raster image densities keep changing. So artifacts like moiré patterns will show up now and again on ALL cameras. It's just mapping one pattern across another. Simple as that.

Only actual SOLUTION is to art direct avoiding patterns in ALL your shots - and good luck with that!

[Jon Fairhurst]

The Johnny Carson moire wasn't necessarily aliasing. It was "chroma crawl". Any frequencies in luma near the chroma subcarrier would be interpreted as chroma, which would roll through that part of the image. Conversely, high frequency chroma would stray from subcarrier and would be seen as moving dots of luma, or "dot crawl."

Improved decoders implemented 2D filters to separate luma and chroma. Carson's herringbone jacket would still mess it up though. As digital processing improved and memory prices came down, 3D filters were employed, which work great until things move - and jackets on people don't sit still.

High end encoders were also developed and deployed. They were smart enough to throw away the offending frequencies. In later Carson episodes, you would see solid gray patches on the jacket, rather than crawling colors.

When at Grass Valley Group, I was an engineer on the Emphasys encoder and decoder line, which used 2D and adaptive techniques. So, which company bought the first Emphasys encoder? QVC! Image quality was more important to them than the networks. They wanted those diamonds that they were hawking to sparkle, and we had some adaptive high frequency boost algorithms that cranked up the glint - without turning into chroma crawl. :)

[Jon Fairhurst]

Hey, I found a photo. :)

http://www.marconitelecine.com/Graph...s/P1010029.jpg

[Bill Pryor]

I see moire on occasion in major TV shows.

[Bill Davis]

Originally Posted by Jon Fairhurst

Hey, I found a photo. :)

http://www.marconitelecine.com/Graph...s/P1010029.jpg

I happily bow to John's superior technical knowledge.

Still, I stand by the fact that no matter how cameras evolve, there are going to be resolutions limiits in terms of both horizontal and vertical scanning grids, and if that conflicts with picture content - brick patterns on the building, window grids OR tie stripes in wardrobe, I bet we'll still have to make small compromises on our framing and shots in order to change the way the pattern in life hits the sensor in order not to generate artifacts.

Though I'd LOVE to see a camera that could handle this in the way that CLEAR SCAN technology handled the raster offset of video screen timing in pre-LCD screen times gone by.

[Jon Fairhurst]

Originally Posted by Bill Pryor

I see moire on occasion in major TV shows.

Yep. Aliasing definitely happens.

In fact, you will probably see it on ABC, FOX, and ESPN, if you have a 1080p TV, and on the other networks if you have a 720p TV. The upscaling and downscaling quality in TVs and STBs isn't always stellar.

In any case, it is usually a compromise between aliasing and softness. (Not so much on DSLRs. They just plain alias.) You can shoot and display with no aliasing (in theory), but that usually gives a softer picture than if you allow a bit of aliasing.

The way to see if aliasing has been avoided is to look for high frequency areas with gray patches. The gray areas are where the high frequencies (that would alias) were removed.

[Jeff Murray]

Michel Coleman kindly replied to my question on vimeo. We were right it was shot on a 5D and 7D. Is it just me or is there a noticeable look from the DSLR, or even the Canon EOS?

"Michael Coleman 1 day ago
Hey Jeff! We shot these interviews on a 5D and 7D..."

[Ben Denham]

A few week's ago I was watching "Men who stare at goats" on DVD. I think it was being up-scaled by the HDTV that I was watching it on. There's a scene in there where the George Clooney character is stumbling through a sandy desert. It cuts to an elevated-wide-shot and at that point all the little ridge-lines in the sand started moire-ing around all over the shop. I thought to myself "how could they even consider using that shot?" but going from what Jon says perhaps it was an artefact of the up-scaling.

[Jeff Murray]

You are right ont the upscaling, I did say 'or some other production/workflow issue'. It may even be the flash container on Vimeo. I do find a lot of the moiré in my 5D videos is dealt with when I them take back to mp4 from the Avid OMF files.

With regards

Jeff

[Daniel Browning]

Originally Posted by Jon Fairhurst

The upscaling and downscaling quality in TVs and STBs isn't always stellar.

Understatement is the soul of criticism. :)

Originally Posted by Jon Fairhurst

In any case, it is usually a compromise between aliasing and softness.

Don't forget ringing. Even if the resampling doesn't add ringing by itself, you can always do a separate sharpening pass to make sure the ringing artifacts get added in.

Personally, I prefer softness over artifacts.